Benchmark Almanac

Clipping Bench

GPT-6 Astra $9,514 leads since Oct 2026 Not in model cards

Agents run a simulated clip farm for about a month. They join pay-per-view campaigns, cut clips from licensed streams, post them and compete with rival clippers for budgets.

Scored by cash at the end of the simulated season. 19 models in the Fireworks index.

Made by Naïve Released Oct 2026 Simulated companies Tasks kept private Last model added Sep 2026 Scores refreshed 2 Nov 2026
Level
Expert Specialists with years of training, often over hours of work
Scored
Settled by events Predictions are scored against what later happens.
Answers
Agent task Many steps in a terminal, browser, desktop, app, game or robot.
Leaderboard usenaive.ai
19
Models scored on its own leaderboard

GPT-6 Astra leads with $9,514

Best score per model from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best.

19 models. Scroll sideways for the rest.

  1. $9,514 GPT-6 Astra
  2. $5,707 GPT-6.1 Sol
  3. $2,018 Claude Opus 5.5
  4. $1,240 GPT-6 Sol
  5. $1,068 Claude Fable 5.1
  6. $886 Claude Sonnet 5.5
  7. $805 Grok 4.6
  8. $734 Claude Opus 5
  9. $644 GLM-5.3
  10. $596 Gemini 3.8 Flash
  11. $462 DeepSeek-V4-Pro
  12. $458 Qwen 3.6 Plus
  13. $399 GLM-5.3-Flash
  14. $384 Grok 4.7
  15. $374 GPT-6 Luna
  16. $343 Kimi K3
  17. $320 DeepSeek-V4-Flash
  18. $223 MiniMax-M3
  19. $218 Gemini 3.5 Flash
  1. 1 GPT-6 Astra $9,514
  2. 2 GPT-6.1 Sol $5,707
  3. 3 Claude Opus 5.5 $2,018
  4. 4 GPT-6 Sol $1,240
  5. 5 Claude Fable 5.1 $1,068
  6. 6 Claude Sonnet 5.5 $886
  7. 7 Grok 4.6 $805
  8. 8 Claude Opus 5 $734
  9. 9 GLM-5.3 $644
  10. 10 Gemini 3.8 Flash $596
  11. 11 DeepSeek-V4-Pro $462
  12. 12 Qwen 3.6 Plus $458
  13. 13 GLM-5.3-Flash $399
  14. 14 Grok 4.7 $384
  15. 15 GPT-6 Luna $374
  16. 16 Kimi K3 $343
  17. 17 DeepSeek-V4-Flash $320
  18. 18 MiniMax-M3 $223
  19. 19 Gemini 3.5 Flash $218
#ModelScoreSettingDateSource
1 GPT-6 Astra $9,514 Sep 2026 Official leaderboard
2 GPT-6.1 Sol $5,707 Sep 2026 Official leaderboard
3 Claude Opus 5.5 $2,018 Sep 2026 Official leaderboard
4 GPT-6 Sol $1,240 Sep 2026 Official leaderboard
5 Claude Fable 5.1 $1,068 Sep 2026 Official leaderboard
6 Claude Sonnet 5.5 $886 Sep 2026 Official leaderboard
7 Grok 4.6 $805 Aug 2026 Official leaderboard
8 Claude Opus 5 $734 Jul 2026 Official leaderboard
9 GLM-5.3 $644 Aug 2026 Official leaderboard
10 Gemini 3.8 Flash $596 Sep 2026 Official leaderboard
11 DeepSeek-V4-Pro $462 Apr 2026 Official leaderboard
12 Qwen 3.6 Plus $458 Mar 2026 Official leaderboard
13 GLM-5.3-Flash $399 Aug 2026 Official leaderboard
14 Grok 4.7 $384 Sep 2026 Official leaderboard
15 GPT-6 Luna $374 Sep 2026 Official leaderboard
16 Kimi K3 $343 Jul 2026 Official leaderboard
17 DeepSeek-V4-Flash $320 Apr 2026 Official leaderboard
18 MiniMax-M3 $223 Jun 2026 Official leaderboard
19 Gemini 3.5 Flash $218 May 2026 Official leaderboard

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Own leaderboard 1 leader
GPT-6 Astra $9,514

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard
$0 $2,000 $4,000 $6,000 $8,000 $10,000 2026 $0 $2,000 $4,000 $6,000 $8,000 $10,000 2026

No lab model card reports it yet

We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.

More in Economic work

All 91 in Economic work
  • JobBench In use 130 tasks from 35 occupations workers most want delegated. Agent works through messy reference files, graded by fact-anchored binary rubrics. GLM-5.3 61.4%
  • GDPval In use Real work tasks from 44 occupations with reference files. Experts blindly compare model deliverables with expert work. Metric is win-or-tie rate. Seed2.1 Pro 87.9%
  • GDPval-AA v2.1 In use Agents with shell and web access complete 220 GDPval gold tasks from 44 occupations. Blind pairwise comparisons give Elo ratings. Claude Opus 5.5 1,846
  • AA-Briefcase v1.1 In use Version 1.1 of Artificial Analysis's multi-hour office work evaluation, with Elo ratings fitted by Crowd-BT. Claude Opus 5.5 1,822
  • CoWorkBench In use Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished). Qwen3.8-Omni-Flash 75.3%
  • OfficeQA In use Grounded reasoning over a large corpus of U.S. Treasury Bulletin documents: locate tables and compute precise numeric answers. Claude Opus 4.7 86.3%